pair
image
point
window
ones
way.
ion
If the
red as
ference
t lower
rch
nding
lower
|: track-
meas-
mber of
> block
ively.
ound in
nent
arge
s could
of fully
areas
ts either
1e block
na
the
et lost
asure-
lits.
hey have
allows
addition,
lows are
nt
nts
), a
DIS AT.
in
yrithm
ccurate
e to be
| be
dint
de of
win-
GCP
initial
measurement, or click the new point in the point list if it
was measured in a measurement window before,
* Preposition the point in the window, where an automatic
prepositioning may also be possible (cf. coming up texts),
e In the case of manual measurement, digitize the point in all
windows in which it occurs,
e In the case of semiautomatic measurement, digitize the
point precisely in one window and approximately in the
other ones, where the real measurement is then performed
by ISM which is able to reach an accuracy of up to 0.1
pixel size,
e In the case of automatic measurement, digitize the point
precisely in one window only, after that the point is trans-
formed to other corresponding windows by means of the
exterior orientation parameters and automatically measured
by a Multi-ISM, which simultaneously matches the points
in all measurement windows. This mode is especially to
recommend after the fully automatic tie point determina-
tion or after a first block adjustment.
Since GCP measurement can be very time-consuming, in par-
ticular if there are many GCPs to be measured, PHODIS AT
offers a further option:
* Measure only as many GCPs as required for a first block
adjustment,
e Perform the block adjustment,
* Read the exterior orientation parameters of images into
PHODIS AT,
* Then the operator is guided to those GCPs that have not yet
been measured so that he can measure them as described
above.
This method can be used for the whole block or the subblocks,
and offers the following advantages:
* The operator is assisted in the search for GCPs.
* Misinterpretation of GCPs is virtually excluded, and the
operator does not have to enter point names. This is a
common source of errors in conventional aerotriangulation
measurement, and their recovery is often very time-
consuming.
4 Block adjustment
PHODIS AT intentionally does not include an own block ad-
justment program. The advantage is that customer-owned block
adjustment programs can be linked to PHODIS AT and that the
operator does not need to give up his/her familiar system.
Interfaces can easily be created to read and write data between
these systems and PHODIS AT. If the adjustment program runs
on the same platform as that of PHODIS AT, it can be started
directly from the environment of PHODIS AT. Otherwise, the
block adjustment program can be started on any other comput-
ers.
Adjustment can be done either for the whole block or the sub-
blocks.
S Block post-processing
5.1 Creation of control and new point sketches
PHODIS AT contains a module for the creation of control and
new point sketches in the Postscript format. The appearance of
these sketches can be varied by the user. They essentially
contain the following elements:
e Overview section from the image file showing the location
of the point in the image (coarse image location).
* Detail section of the image showing the precise location of
the point in the image.
* Explanatory information such as the name, coordinate
values, image file, PHODIS project etc.
5.2 Data editing for further use
The purpose of aerotriangulation is the acquisition of the exte-
rior orientation parameters of the images or the relative and
absolute orientation parameters of individual stereo models.
PHODIS AT therefore offers the option to read in the data
directly into PHODIS and then to use it without further proc-
essing in other PHODIS projects, and to create PHOREX or
PEX files which can then be read directly into analytical sys-
tems from Carl Zeiss.
During this process, stereo models are formed and the exterior
orientation parameters of the images are split up in relative and
absolute orientation parameters of the models.
Because of the very strong bridging of the images, parallaxes
are minimized optimally in the models. Thus, a new relative or
absolute orientation in an analytical plotter is neither required
nor recommended. What an operator can do for orientation is
only the model checking. Since it is not necessary to measure
GCPs or new points in every model during the block measure-
ment, tie points determined by the fully automatic measure-
ment can be used for model checking. The quality of a model
in an analytical plotter can be checked as follows:
* A planimetric check can be achieved by setting existing
GCPs or new points. If there are none of GCPs or new
points available, those tie points that are suited for a
planimetirc check can be used.
e Elevation checking can be performed by using any GCP,
new or tie point.
6 Practical results
Four blocks were used to test the reliability and accuracy of
aerotriangulation by PHODIS AT. These blocks meet the fol-
lowing requirements: different image resolutions, different
image scales, different camera focal lengths and different
image materials.
Block Forssa:
This block is the OEEPE block provided for the OEEPE test
"Digital Methods in Aerial Triangulation". The block consists
of 28 images with 60-70% forward and 15-40% side overlap.
The images show a flat, urban and industrial area.
Block Moeck:
The block consists of 15 images with ca. 60% forward and 20-
30% side overlap. The images present a rural, woody and hilly
landscape.
Block Graz:
This block is an inner city flight with a long focal-length cam-
era. Color images with ca. 60% forward and 30% side operlap
show a varied city area.
Block Echallens:
This block is featured by high forward (ca. 80%) and side
overlaps (ca. 60%), and presents a rural landscape.
The results of the measurements and adjustments achieved are
shown in Table 3. None of the 4 blocks required interactive
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
